Use of uninitialised data in tc-sh.c

[Andrew STUBBS]

> I think this solves it.  At least, it looks right, passes valgrind,
> and passes make check.  Note that, for each case where this code
> accidentally failed before, it ends up doing the right thing anyway.
> (that's not an excuse, just noting this is a low priority patch)

I have tested your patch and it does indeed pass both valgrind and plumhall.

I assume you have decided not to worry about the implicit long long to long
cast in the assignment of val.

--
Andrew Stubbs
andrew.stubbs@st.com
andrew.stubbs@superh.com